Re: Atom protocol and digital signatures

[James M Snell <jasnell@xxxxxxxxx>]

The main point of this discussion is that what happens is completely up
to the server implementation.  It is entirely possible for a particular
server to do exactly what you are suggesting.  However, it is impossible
for a client to make any assumptions about how a server will treat
signed entries without prior specific knowledge of that server.

- James

Bob Wyman wrote:
> 
> I think it should be remembered that signed Atom entries *do* make a
> great deal of sense within the context of the Atom formatted documents
> -- whatever difficulties that they may cause in the protocol
> discussion.
> 
> One important use-case discussed during the definition of the Atom
> syntax was that of aggregate or synthetic feeds. (i.e. feeds that are
> composed of entries taken from more than one other feed).  The current
> syntax allows one to preserve a signed entry in an aggregate feed --
> as long as the signed entry had an atom:source element when it was
> originally signed.
> 
> If the protocol can't preserve the signature associated with an entry
> provided by a client, then it would seem that the protocol can't be
> used to construct an aggregated feed that preserves the integrity of
> properly signed atom entries extracted from other feeds. That is
> unfortunate and seems extremely non-optimal.
> 
> bob wyman
> 
>